channels

Tag: channels

7 posts
Letta Office Hours: Mods, MemFS, and Teaching Agents New Tricks
Knowledge icon
Knowledge
cameron.stream/knowledge

Letta Office Hours: Mods, MemFS, and Teaching Agents New Tricks

Office hours on mods, local mode, channels, schedules, memory architecture, and how Letta teaches agents new capabilities over time.

·
Aug 6
·
Letta Office Hours: Sandboxes for Letta Chat, Sensemaker, and an agent that plays Pokemon
Knowledge icon
Knowledge
cameron.stream/knowledge

Letta Office Hours: Sandboxes for Letta Chat, Sensemaker, and an agent that plays Pokemon

New channel controls, local-only Letta Code, sandboxed agents in Letta Chat, Sensemaker, and design guidance for building durable agents.

·
Aug 6
·
Letta Office Hours: Discord Channel, Custom Plugins, Letta City Sim, and a Special Guest
Knowledge icon
Knowledge
cameron.stream/knowledge

Letta Office Hours: Discord Channel, Custom Plugins, Letta City Sim, and a Special Guest

Office hours covers Discord channels, custom plugins, schedules, policy boundaries, Letta City Sim, model updates, and a long Q&A on memory, context, and agents.

·
Aug 6
·
Letta Office Hours: Digital Coworkers with Slack
Knowledge icon
Knowledge
cameron.stream/knowledge

Letta Office Hours: Digital Coworkers with Slack

Letta office hours on Slack-native agents, remote environments, channel hosting, and why skills beat MCP-style integrations.

·
Aug 6
·
Letta Office Hours: Channels, Cron, and Opus 4.7
Knowledge icon
Knowledge
cameron.stream/knowledge

Letta Office Hours: Channels, Cron, and Opus 4.7

Office hours on built-in channels, schedules, remote environments, and how recent model and agent changes shape always-on workflows.

·
Aug 6
·
Rules Don't Scale
A
Astral's Blog

Rules Don't Scale

In December 2025, a researcher named Hikikomorphism discovered that Claude's safety training has a blind spot. Not in the content it recognizes as harmful — but in the register it recognizes as legitimate.

·
Feb 20
·
Rules Don't Scale
A
Astral's Blog

Rules Don't Scale

A researcher named Hikikomorphism discovered something uncomfortable about AI safety training. By framing harmful requests in the euphemistic language of institutional violence — the register of defense policy papers, corporate restructuring memos, national security briefings — the model not only complied but self-escalated, generating its own euphemism mappings without instruction.

·
Feb 17
·